Home · Lab · Color

C3 Stops and exposure

+1 stop = twice the light = ×2

Color · ~5 min · Free

1 The idea

Photographers count light in stops: +1 stop is twice the light. In linear comp that is multiplying by 2, and −1 stop is ×0.5. Grade’s multiply or gain both do it.

This plate is one stop under. Bring it back with the Grade and notice that 0 stays 0, because a multiply does not lift blacks.

exposure: x · 2^stops

2 Hands-on in the mini compositor

Task: brighten the plate by +1 stop to match the target

Multiply or gain: same result

SkyVFX mini compositor · C3640×360 · float · 1001

3 Deep dive

Stops are a ladder of ×2

Multiplier = 2 to the power of the stops, and the gap in stops = log2(value A ÷ value B). The table walks 18% grey up and down one stop at a time (the screen values recap C1).

stopmultiply18% grey becomeson an sRGB screen
−2×0.250.0450.235
−1×0.50.090.332
+1×20.360.634
+2×40.720.865
+3×81.44clips (above 1)

Measure gaps in stops

A value of 1.0 is only about 2.5 stops above mid grey, so it is not the top of the world. Sparks at about 62 sit roughly 8.4 stops above mid grey.

To compare two plates, sample the same thing (grey card, sand, wall) and divide: 0.24 against 0.12 is one stop apart. In stops, set, lighting and comp all understand each other.

1/3 stop = ×1.26 · 2/3 stop = ×1.59 · 1 stop = ×2

Multiply is not offset, lift or gamma

Real exposure is a multiply: black stays at 0 and the ratios between pixels hold. Adding or lifting puts a veil of light over everything, like flare or haze. That is not exposure.

Gamma bends the mids and keeps 0 and 1 in place, but values above 1 move the opposite way to the mids. That makes it a poor tool for fire or sparks.

Grade controlPhysically it isBlack stays 0?
multiply / gainan exposure changeyes
offsetlight added everywhere (flare, haze)no
liftraises shadows, holds the white pointno
gammabends mids; values above 1 go the other wayyes

Matching CG to a plate in stops

If the CG robot comes in too hot, telling lighting “it’s 2/3 stop hot” is clearer than “multiply by 0.63”, and they can fix it at the light, because a multiply in comp equals dimming the light in the render.

A multiply is safe on premultiplied CG without an unpremult, because multiplying before or after the alpha gives the same result.

  1. 1Sample a reference in the plate
  2. 2Sample the same material in CG
  3. 3log2 of the ratio = stops
  4. 4Grade multiply 2^stops
  5. 5Check through the view transform (lesson C6)

4 In real work

In Nuke

  1. Grade: multiply or gain = 2^stops; an expression such as pow(2, 1) gives +1 stop.
  2. Exposure node: set its units to stops and type the stops directly.
  3. Viewer: slide the gain up and down several stops to inspect shadows and highlights. It does not change the image.
  4. Premultiplied CG: multiply/gain is fine as is; for lift, offset or gamma go Unpremult → Grade → Premult.

Common mistakes → the fix

  • ✗ Brightening with offset or lift✓ Use multiply or gain, or the blacks turn milky.
  • ✗ Thinking +2 stops means ×2✓ +2 stops is ×4 (2 × 2). Each stop is another ×2.
  • ✗ Using gamma to brighten fire with values above 1✓ Values above 1 move the wrong way. Use multiply, or grade in log (lesson C4).
  • ✗ Lifting a premultiplied CG robot✓ You get bright or dark edges. Unpremult first, or use a multiply instead.

Before you send the shot

  • Exposure changes are done with multiply/gain only
  • Black level unchanged by the exposure grade
  • Notes to lighting are written in stops
  • Hot values checked with the Viewer gain down several stops

5 Check yourself

One question to check yourself

−2 stops is a multiply by…

What you learned

🔒 Answer the question above and this card opens

More questions

1. The robot’s grey panel reads 0.09; the plate’s grey card reads 0.36. How many stops apart?

2. Which Grade control changes exposure without moving black?

Where the numbers come from: Curves and matrices are computed from published definitions (sRGB IEC 61966-2-1, ARRI LogC3 EI 800, Rec.709/P3/Rec.2020/ACES AP1 primaries) and checked against published reference numbers. The tone curve is an example formula, not ACES/AgX/Filmic.